LILT’s Post

Discover how both DeepL and Lilt accommodate industry-specific terms through glossaries, with Lilt offering additional segmentation for more contextualized responses. Learn how accurate translation can vary based on context, making customization crucial. Watch the short webinar clip on YouTube: #ai #GenerativeAI #EnterpriseAI #Localization #Globalization #l10n #i18n

Worth every minute of it. Cast about the BS spun around AI and language models with Timnit Gebru and Emily Bender. A few highlights: -       LLMs are not Artificial Intelligence. One of the reason is that: -       There is no agreement what “intelligence” is in the first place. -       LLMs are “stochastic parrots” spouting words that match strings of words seen most frequently on internet. -       Applications like ChatGPT can still be dangerous for individual users and entire societies alike, and have questionable legal origins. -       The famous letter to “pause the AI research for 6 months” was a marketing play to actually promote AI. “Look how dangerously advanced our AI is becoming, we are so smart”. (I love this take but I hope not all signatories thought that way) - Same goes for Sam Altman calling for AI regulation. Marketing of their product and self-promotion. https://lnkd.in/gnqhj_PA

Adam Conover talks with Emily Bender and Timnit Gebru in a video from 3 months ago that still feels relevant to me today. These two uber-qualfiied ladies express my skepticism over the current AI hype more eloquently than I could do. Current AI systems are stochastic parrots, hence his guests, two of the authors of the famous paper on "The Dangers of Stochastic Parrots". The TL;DR: on the paper is that ChatGPT and other tools based off of Large Lange Models (LLMs) are merely able to "create a convincing imitation of a piece of text that will most readily convince someone who knows nothing about the field". Or, even simpler: LLM bots are con artists and convincing liars, but little more. Adam gives an example of why real world context matters so much: why it is very expensive to have an actor get wet on camera. (Watch at the 11 minute mark.) His short anecdote reminds me of all the context that we needed to unravel to make a high quality translation when I managed and led the translation team for a volunteer-run video game translation project. Context is king and, at least for now, humans are significantly better at digging in to unspoken or missing context. Computers seem oblivious to the fact that anything could be missing in the first place, because that's not what they are designed to do. Check out the video, as well as the others in Adam's series on the honest state of AI - without the hype. https://lnkd.in/gsPUWngW

Got AI FOMO? Wondering what is and isn't going to happen at this week's AI Safety Summit? For the latest, check out our blog and short explainers at https://lnkd.in/ebnDf9Xc, with insights from Freshfields Bruckhaus Deringer's #AI experts across the world. #AI #ArtificialIntelligence #AISafetySummit

Binary Classification: Understanding Activation and Loss Functions with a PyTorch Example ... https://lnkd.in/ehFS8RB9 #AI #ML #Automation

"🌟 Dive into the world of Artificial Intelligence with my latest video ‘AI for All’! 🤖✨ Discover how AI is transforming every aspect of our lives, making technology more accessible and empowering individuals across the globe. From simplifying daily tasks to solving complex problems, AI is the key to unlocking a world of possibilities. Don’t miss out on this insightful journey into the future – watch now and join the conversation about how AI is shaping our world for the better. #AIForAll #TechTalk #FutureIsNow

The Digital Socrates: Leveraging LLMs For An Introspective Journey Into Information, Knowledge, And Wisdom Establishing the new domain of cognitive intimacy. #LLMs #AI #GPT3 #GPT4 #ChatGPT #introspection #cognition

New video 🚀 Exploring pre-quantized LLMs, compression methods, and the package for loading them! This video will explore quantization methods like GPTQ, GGUF (formerly GGML), and AWQ. More specifically, we will explore several quantized models and the packages that help you leverage these compression methods. Hopefully, this helps those deciding which method to use and the techniques/packages to use them! #datascience #ai #machinelearning

Anthropic Releases Claude 3.5 Sonnet, Beats GPT-4o on Benchmarks Things just keep getting better and better! Claude 3.5 Opus, will be released soon with features such as **web search**!! https://lnkd.in/gFjsKgVp #genai #change #disruption

Code Dependent by Madhumita Murgia review understanding the human impacts of AI Daunting tales of how everyday algorithms are changing usWhat’s in the box? That’s the question almost everyone that Madhumita Murgia speaks to seems to be asking. If the “black box” of the algorithm is going to make critical decisions about our hea... https://lnkd.in/eE67Z-D9 #AI #ML #Automation